May 8, 2026 · 8 min read · E-commerce, KPI, Monitoring Plan
KPI monitoring guide for e-commerce companies
In e-commerce, conversion is everything — and conversion can break silently. A checkout button that stopped working at 11pm, a payment processor that started declining 8% more cards than normal, a promotional code that applied discount logic to the wrong SKUs. These events happen constantly, and they cost real money for every hour they go undetected.
The good news is that e-commerce companies tend to have their transaction data in a warehouse already. The problem is that nobody's watching it continuously. Someone's checking a dashboard in the morning, maybe. But the checkout that broke at 11pm isn't discovered until the 9am standup — after 10 hours of lost revenue.
This guide covers what to monitor in e-commerce, at what frequency, and with what comparison logic to get alerts that mean something.
Why checkout conversion deserves 15-minute monitoring
The checkout funnel is where your revenue actually happens, and it's where things break most expensively.
Payment processors have incidents. A/B tests have bugs. Third-party scripts time out and block form submission. Discount code systems apply incorrect logic after a bad deploy. A new address validation library rejects valid postal codes in a specific region.
Every one of these shows up immediately in checkout conversion — and none of them show up in infrastructure monitoring. Your servers are healthy. Your API is responding. Your error rate is normal. But somewhere in the funnel, real customers are hitting a wall and leaving.
15-minute monitoring of checkout conversion is the difference between a 15-minute revenue impact and a 10-hour one.
The e-commerce monitoring plan
| Metric | Frequency | Compare Period | Alert Threshold | Why It Matters |
|---|---|---|---|---|
| Payment success rate | Every 15 min | Rolling 24h average | Drop >2 percentage points | Every failed payment is a lost order — this is the highest-priority metric in e-commerce |
| Checkout-to-purchase conversion | Every 15 min | Same 15-min slot, same day last week | Drop >10% | The most direct signal of checkout functionality — a drop here means something broke |
| Cart-to-checkout conversion | Hourly | Same hour, same day last week | Drop >15% | Catches issues in the cart and early checkout stages |
| Order volume | Hourly | Same hour, same day last week | Drop >25% | Primary revenue activity signal |
| Revenue (gross) | Hourly | Same hour, same day last week | Drop >20% | Top-line revenue health |
| Average order value | Hourly | Same hour, same day last week | Drop >15% or spike >30% | A drop can signal a discount issue; a spike can mean a pricing error |
| Session volume | Hourly | Same hour, same day last week | Drop >30% | Traffic source health — a drop here means acquisition is broken, not conversion |
| Return rate | Daily | Same day last week | Spike >30% relative | Product quality or description accuracy issue |
| Revenue by category | Daily | Same day last week | Drop >25% | Category-level health check — catches issues specific to a product line |
| Promo/discount redemption rate | Daily | Same day last week | Spike >100% relative | Promo code leak or misconfiguration signal |
| Refund rate | Daily | 7-day rolling average | Spike >50% relative | Quality or fulfillment issue signal |
The difference between session, cart, checkout, and purchase metrics
Each stage of the funnel can break independently. Monitoring only order volume is like checking engine temperature but not oil pressure — you'll catch a total failure, but not the quiet problem that becomes a big one.
Sessions → cart add: measures whether users are engaging with products. A drop here means your traffic is lower quality, your site is slower, or a product listing broke.
Cart → checkout initiation: measures whether users are getting stuck before they try to buy. Drop here often indicates a cart UX issue, a persistent login problem, or a broken coupon field.
Checkout → purchase: measures whether the purchase step itself is working. This is where payment processor issues, form validation bugs, and address lookup failures show up. Monitor this every 15 minutes.
Purchase → fulfillment: measures whether orders are processing correctly after payment. A drop here (visible as a spike in failed order processing) indicates a fulfillment or inventory system issue.
Key patterns in e-commerce
Weekdays vs. weekends differ significantly by category. Grocery and home goods peak on weekends. B2B and office supplies peak on weekdays. Fashion varies by season. Never compare a Saturday to a Tuesday. Same-day-of-week is non-negotiable in e-commerce.
Promotional events inflate baselines. A Black Friday or flash sale creates a revenue peak that, when it ends, looks like a catastrophic drop. Your monitoring system needs to handle this — either through baseline suppression during events, or by using a longer rolling average that naturally absorbs spikes.
Payment success rate has a time-of-day pattern. Fraud detection systems are more aggressive at off-hours, which can cause higher decline rates at night. Your payment success rate at 3am will naturally be lower than at 3pm. Use same-time-last-week comparisons, not absolute thresholds, for this metric.
End of month matters for B2B e-commerce. Corporate purchasing spikes at end of month as teams use budget. An end-of-month volume spike followed by an early-next-month dip is normal — not an alert.
Geographic conversion differences are real. Payment success rate can differ 10-15% between regions due to local card issuer policies. If you operate internationally, monitor payment success rate by region so a UK processor issue doesn't hide behind healthy US performance.
A note on campaign traffic spikes
Paid traffic campaigns create temporary session and order volume increases that distort your baselines. When the campaign ends, volume returns to normal — and your monitoring system will alert on the normalization if you're using a short comparison window.
The solution: use a 4-week same-day-of-week average as your baseline, not just last week. One campaign week doesn't shift a 4-week average dramatically, so normalization after a campaign doesn't trigger false alerts.
What a good e-commerce alert looks like
A checkout conversion alert:
S1 — Checkout conversion drop Checkout-to-purchase rate: 67.3% — down 14.1 points vs same 15-min window last Tuesday (81.4%). Period: 11:15–11:30pm · Segment: US · Device: All [Acknowledge] [Escalate] [Mark as known issue]
A payment success rate alert:
S1 — Payment success rate below baseline Payment success rate: 93.8% — down 3.2 points from 24h rolling average (97.0%). Period: last 15 min · Processor: Stripe · Segment: All regions [Acknowledge] [Escalate]
These are actionable because they tell you exactly what dropped, by how much, compared to what, and in which segment. Without the segment context (US, Stripe), you're just guessing where to look.
Starting point for new setups
Start with three metrics:
- ·Payment success rate — 15-minute monitoring, rolling 24h average. Nothing is higher priority.
- ·Checkout-to-purchase conversion — 15-minute monitoring, same window last week.
- ·Hourly order volume — same hour last week. Your primary revenue pulse.
Once those are tuned and you're confident in the baselines, add the rest of the funnel above it (cart-to-checkout, session volume) and below it (refund rate, return rate).
Lighthouse connects to your data warehouse and monitors e-commerce metrics continuously, with Slack alerts when they move. Start for free →